We just shipped sunpeak.ai as a standalone testing framework for MCP Apps!
If you're building MCP Apps for ChatGPT or Claude, you know the pain: deploy, open the host, start a conversation, trigger the tool, check the result. Repeat for both hosts, both themes, three display modes. That's 24 combinations per code change.
sunpeak replicates the ChatGPT and Claude runtimes locally. You write Playwright tests that call tools, render resources, and assert against the output. One test file runs against both hosts automatically.
What's included:
- Unit tests (Vitest + happy-dom)
- E2E tests against replicated ChatGPT and Claude runtimes
- Visual regression testing with screenshot baselines
- Live tests against real ChatGPT
- Works with any MCP server in any language (Python, Go, TypeScript)
Add it to an existing project with one command:
pnpm add -g sunpeak && sunpeak test init
No paid host accounts. No AI credits. Runs in CI/CD.
MIT licensed and open source! https://sunpeak.ai/testing-framework/
Top comments (0)